Dear Jacob,

Thanks for that explanation. Does this mean that eddy_openmp needs to have more than a single b=0 volume? In the data I have shown, only the first volume was b=0, the rest were DWIs at b=1000. By your explanation, eddy_openmp seems to be saying that all my DWIs are co-registered, but not to the b=0 volume. I have visually inspected the data, and this is not the case - there is motion between DWI volumes. Did I misunderstand your explanation?

well, yes I think maybe so. One does not need two b=0 images at the beginning. It was just an attempt to understand why there are non-zero values between volume 1 and the rest, but zero values between all other volumes.


What I find strange is that 'eddy' from fsl/5.0 writes different movement parameters and non-zero eddy currents for each volume (100% of the time),  while  'eddy_openmp'  writes repeated values and zeroes for all the volumes (in 40% of our dataset). 

As I said, something is clearly wrong. But it is impossible to say what from the information I have.


Looking at the post-processed NIFTI images, it is apparent that motion correction has happened using eddy_openmp. But because of the way these files are being written, it is hard to trust that what I'm eyeballing is correct. 

Yes, you should not use these data until we know what is going on. But it sound like maybe the problem is not in the estimation of the parameters, but rather between applying them and writing them to a file.


Yes, we will wait for the fsl/6.0.1 release, at which time I will write to this thread again. Do we have an estimate on when that release would be?  

We are aiming for before Xmas.

Jesper


Regards,
Jacob




On Fri, Nov 30, 2018 at 5:36 AM Jesper Andersson <[log in to unmask]> wrote:
Dear Jacob,

>
> We've been using fsl/5.0 for a while, but we wanted motion parameters from the eddy_movement_rms files. To that end, I'm now using eddy_openmp from fsl/5.0.11.
>
> While eddy_openmp seems to clearly be returning motion corrected NIFTIs, the actual eddy_movement_rms files looked incorrect, because they were mostly populated with zeroes and repeated values, for ~40% of our dataset. These jobs were all multi-threaded (pe threaded 4).
>
> Example of one of our 'bad' eddy_corrected_data.eddy_movement_rms file;
> 0  0
> 0.2097059397  0.2097059397
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
> 0.2097059397  0
>
>
> When we check the "eddy_corrected_data.eddy_parameters" file, it shows us this;
> 0  0  0  0  -0  0  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
> 0.1028521901  0.3712571436  -0.1851838779  -0.006534020416  -0.0004291008227  -0.004189225845  0  0  0  0  0  0  0  0  0  0
>
>

as you say this is clearly not right. The way it works internally is that it first does a rigid body alignment of all b=0 volume, does a linear interpolation of movement parameters between the b=0 volumes and uses those as staring estimates for the DWIs. In this case, assuming that the two first volumes are b=0, it seems that the DWIs go nowhere (the same movement parameters as the b=0 and no eddy currents).

I will be very happy to take a look at your data as soon as I have the 6.0.1 release out of the way. Please send me an email and remind me after we have announced that. I don’t think it is meaningful to look at it before then since there is a decent chance that 6.0.1 will sort out the problem.

>
> I do not know the inner workings of eddy_openmp, and was hoping the developers or users may be able to help me figure out why we're having so many repeated values and zeroes written to the files. The actual NIFTI image output appears to be okay, but due to these zeroes, I cannot be sure if it is okay to go ahead and use them.

When you look at a movie of the original images and compare that to a movie of the “corrected” images, do you see a reduction in apparent movement/distortions?

If I was you I would be wary of these data until I have had a chance to look at it.

Jesper


>
>
> Hoping to hear from you soon
> Jacob
>
> ########################################################################
>
> To unsubscribe from the FSL list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1



To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1